Diagnostic Accuracy of Tryptase Levels for Pediatric Anaphylaxis: A Case-Control Study

Anaphylaxis is a severe, potentially life-threatening allergic reaction [1]. It is crucial to correctly diagnose cases of anaphylaxis to manage patients appropriately and to provide patients with an epinephrine autoinjector to use promptly, in case of future reactions [2]. The definition of anaphylaxis is currently based on clinical criteria [3]. Given that patients presenting with other conditions (e.g., acute urticaria, asthma exacerbation) may present with similar clinical symptoms, it is difficult to prove an anaphylaxis diagnosis, using laboratory markers. Studies suggest that changes in tryptase levels are sensitive for the diagnosis of anaphylaxis, especially in severe reactions [4‒8]. Indeed, tryptase levels are reported to increase from 15 min to 3 h after anaphylaxis onset [9]. One study proposed that tryptase levels during reaction higher than 2 ng/mL + 1.2* (baseline tryptase levels) can be used as a rule to confirm diagnosis of anaphylaxis (referred to as strategy 1) [10]. This was shown to have a sensitivity of 78% and specificity of 91% [11]. However, it was recently suggested that acute tryptase/baseline tryptase ratio exceeding 1.685 (referred to as strategy 2) is more accurate for differentiating individuals with higher baseline serum tryptase, from anaphylaxis reactions, with a sensitivity and specificity of 94.4% [12]. In this study, we aimed to compare the two strategies to assess whether changes in tryptase levels can be used as a biomarker for the diagnosis of anaphylaxis in children.

We extracted cases from our anaphylaxis registry consisting of children presenting with anaphylaxis at the Montreal Children’s Hospital (MCH) emergency room (ER) [13]. Only those with measurement of a reaction (within 60–120 min of onset of reaction symptoms) and baseline (at least 24 h postreaction onset and up to 6 months later) tryptase level were included. Anaphylaxis was defined as a systemic reaction involving at least 2 organ systems and/or hypotension [14]. We collected data prospectively and retrospectively through a standardized data entry form as previously described [15, 16]. For prospectively recruited patients, the treating ER physician along with a trained member of the research team identified cases of anaphylaxis and, after obtaining consent, completed the data entry form. The remaining cases were identified retrospectively by searching the electronic medical record system for the International Statistical Classification of Diseases and Related Health Problems, 10th Revision codes related to anaphylaxis. Data were collected regarding demographics (age and sex), types of triggers (food, drug, venom, unknown, and others), presence of comorbidities (asthma, food allergy), as well as the regular use of medications such as nonsteroidal anti-inflammatory drugs and β-blockers. The severity of anaphylaxis was classified as mild, moderate, and severe, using a modified Muraro et al. [17] grading system. Mild anaphylaxis included nasal congestion/sneezing, rhinorrhea, throat tightness, mild wheezing, generalized pruritus, flushing, urticaria, angioedema, mild abdominal pain, nausea/vomiting, tachycardia, and anxiety [18]. Moderate anaphylaxis included harsh cough, moderate wheezing, dysphagia, shortness of breath, crampy abdominal pain, repeated vomiting, diarrhea, and lightheadedness. Severe anaphylaxis included cyanosis, respiratory arrest, hypotension, circulatory collapse, dysrhythmia, bowel incontinence, cardiac arrest or serious bradycardia, confusion, and unconsciousness [19].

Total tryptase levels were measured 60–120 min after the onset of symptoms (referred to as reaction levels) at the treating physician’s discretion. All patients with anaphylaxis were subsequently referred for assessment and management at the MCH Pediatric Allergy clinics. In patients who consented to follow-up by the research team, postreaction tryptase levels (at least 24 h after symptoms resolution and up to 6 months later, referred to as baseline levels) were measured regardless of the level during anaphylaxis. Total tryptase level was measured using UniCAP-Tryptase fluoroimmunoassay (Phadia, now Thermo Fisher Scientific, Uppsala, Sweden) by the MCH central laboratory, following the manufacturer’s instructions.

Over a 10-year period, from 2013 to 2023, children (0–18 years) presenting to the pediatric allergy clinics at the MCH were recruited as part of a chronic urticaria (CU) registry and followed prospectively. CU is also a mast cell-mediated disease, but symptoms (hives/angioedema) are limited to the skin. Consenting patients from our CU registry had two tryptase measurements 1–2 months apart. We assessed tryptase variability among individuals with CU without a history of anaphylaxis, to compare specificity of both strategies.

Descriptive statistics were used to present demographic, clinical, and reaction characteristics, epinephrine use and the change between reaction and baseline tryptase. All statistical data were analyzed using R 4.2.3 binary for macOs 10.3 (R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/). Categorical data were presented as percentages and continuous data as a median with interquartile range (IQR). The demographic, clinical and reaction characteristics of all participants presenting with anaphylaxis with and without measurement of tryptase levels were compared to address the potential for selection bias using χ² test for categorical variables and an unpaired two-tailed t test for continuous variables. These characteristics were also compared between those with tryptase level measured both 1–2 h following reaction and at baseline and those that had only measurement 1–2 h following symptoms onset.

Sensitivity was determined by dividing anaphylaxis cases identified by each strategy’s rule (numerator) by the total cases meeting the definition of anaphylaxis according the NIAID criteria [20]. To assess specificity, the number of cases not meeting the diagnosis of anaphylaxis according to each strategy’s rule (numerator) among the CU cases was divided by the total number of CU assessed (denominator).

The McNemar test was computed to compare sensitivities and specificities of strategies. The Delong test was computed to compare the area under curve (AUC) of both strategies based on sensitivities and specificities to assess which strategy is most optimal.

We established our own receiver operating characteristic (ROC) analysis for the ratio between reaction tryptase and baseline tryptase to determine the optimal cutoff value based on our data.

Ethics approval was granted from the McGill University Health Centre Ethics Review Board (IRB: 10-203 Glen). Patients or the public were not involved in the design, or conduct, or reporting, or dissemination plans of our research. Informed written consent was obtained from the participants’ parent, legal guardian, or next of kin to participate in the study.

From April 2011 to November 2023, 3,158 pediatric patients presented with anaphylaxis to the MCH ER. A total of 746 patients were followed prospectively, while 2,420 patients were assessed retrospectively. Among these patients, 255 had reaction tryptase measured within 60–120 min of onset of reaction symptoms, and the median age was 5.2 years (IQR 1.9–12.2) and the median reaction tryptase level was 5.6 (IQR 3.9–9.4) (Table 1).

Eighty-nine patients had tryptase measured during reaction and at baseline. Within these patients, the median age was 5.2 years (IQR 1.5–11.0), 61.8% were males, and 84.3% presented with a food-induced reaction. 59.6% presented with a moderate reaction severity (Table 2).

Between 2013 and 2023, 251 patients were recruited as part of the urticaria registry. The median age at presentation was 9.4 (IQR 4.8–14.3), and 47.8% were male (Table 3). Within this registry, 25 patients had tryptase levels measured at two different occasions, 1–2 months apart.

To evaluate for potential selection bias, patients with and without a reaction tryptase level were compared (Table 1). Patients with reaction tryptase had a significantly higher (p < 0.01) frequency of reactions during exercise, eczema, and epinephrine administration during hospitalization; they had a lower frequency of known food allergy and epinephrine administration outside of hospital.

Patients with both reaction and baseline tryptase levels were compared to those with reaction levels only (Table 2). Patients with both levels had a lower frequency of moderate reactions and a higher mean serum tryptase.

Of the 89 patients with anaphylaxis who had both a reaction and baseline tryptase measured, 48 patients had reaction tryptase exceeding 2 ng/mL + 1.2* (baseline tryptase). The sensitivity for strategy 1 was 53.3%, and the specificity was 95.0% with 1 patient being a false-positive. Forty-nine of 89 patients had a reaction tryptase over baseline tryptase ratio exceeding 1.685. The sensitivity for this second strategy was 54.4%, and the specificity was 85.0% with 3 patients being false positives. Based on the McNemar test, there is no statistically significant difference (p > 0.05) in sensitivities and specificities between the strategies. In severe reactions, the sensitivity for strategy 1 (92.9%) was similar (p > 0.05) to strategy 2 (also 92.9%).

The positive likelihood ratio for strategy 1 was 10.66 compared to 3.63 for strategy 2 (Table 4). The AUC for strategy 1 was significantly higher than strategy 2 (0.69 vs. 0.66, p < 0.04).

We computed a ROC curve (Fig. 1) with anaphylaxis as the dependent variable and the ratio between reaction tryptase and baseline tryptase as the independent variable. The analysis yielded an AUC of 0.776 with an optimal cutoff of 1.264. At this threshold, the sensitivity of the strategy was 69.7% and specificity was 80.0%.

Our findings revealed that strategy 1 performed slightly better than strategy 2 as per the DeLong test. However, using our own dataset, we demonstrated that strategy 2 can be optimized with a cutoff value of 1.264 instead of 1.685. Our findings differed from other studies as a previous study by Mateja et al. [21] assessing strategy 2 demonstrated a sensitivity and specificity of 94.4%. Differences in sensitivity compared to our study could be potentially explained by different clinical characteristics of the population studied and differences in sample sizes. Our study was focused on children and a large proportion had food allergy, whereas the Mateja study was mainly focused on an adult population with less than 10% of reactions being attributable to a food trigger. Furthermore, our study had a larger sample size than the Mateja study (89 anaphylaxis cases vs. 22 anaphylaxis cases, respectively). Both strategies were seen as being more sensitive for severe reactions versus all reactions.

Anaphylaxis diagnosis remains difficult to prove and remains under-recognized, undertreated, and poorly understood partially attributable to lack of reliable biomarkers. Measuring tryptase levels has been an accurate indicator of mast cell degranulation, and hence, the interest in assessing its accuracy in diagnosing anaphylaxis [22]. However, as both tests had a low sensitivity of 55%, a negative result does not reliably exclude a diagnosis of anaphylaxis.

Our study has some potential limitations. First, the sample size for the measurement of both reaction and baseline tryptase was relatively small, representing around 3% of our overall cohort of patients with anaphylaxis, and there was evidence of selection bias. However, it remains the largest sample assessing these tests among pediatric cases of anaphylaxis. Second, our study relies on determination of reaction severity based on parental report of symptoms in a pediatric population usually given by the parents; the report of symptom severity may not always be reliable within that population. Third, the gold standard for diagnosing anaphylaxis was a clinical diagnosis based on the National Institute of Allergy and Infectious Disease and the Food Allergy and Anaphylaxis Network criteria, which is an imperfect gold standard [23].

We conducted the largest study to date to compare two different strategies using temporal trends in tryptase to confirm the diagnosis of anaphylaxis in children. We demonstrated that although both strategies had similar sensitivities and specificities, strategy (1) was found to be slightly superior to strategy (2). Using either of these strategies for anaphylaxis confirmation is plausible, leaving the decision on which strategy to use to the discretion of the treating allergist. Further, we confirm the utility of tryptase for the retrospective diagnosis of anaphylaxis mainly in severe cases.

Anaphylaxis is a severe allergic reaction diagnosed based on clinical criteria. Multiple studies have investigated using tryptase levels for accurate anaphylaxis diagnosis. However, two distinct strategies are currently reported in the literature for using tryptase in this context. In this paper, we demonstrate, using our database of pediatric anaphylactic reactions, that the strategy of a tryptase level during a reaction exceeding 2 ng/mL + 1.2 times the baseline tryptase level is superior. Additionally, we propose our own strategy to optimize anaphylaxis diagnosis using tryptase. Earlier diagnosis of anaphylaxis leads to better patient outcomes and helps prevent mortality and complications.

This study was approved by the McGill Ethics Board with IRB: 10-203 Glen. Written informed consent was obtained for participation in this study by the patient’s parents, legal guardian, or next of kin.

The authors do not have any conflicts of interest to declare.

The authors have no personal, financial, or institutional interest in any drugs, materials, or devices described in this article.

R.K., C.P., and M.B.S. wrote the manuscript. M.B.-S., C.M., A.B., M.K., and A.E.C. helped gather the dataset and reviewed the manuscript. M.B.-S. was the principal investigator for this study.

Edited by: H.-U. Simon, Bern.

Data involved in this study are not publicly available as their containing information could compromise the privacy of research participants. Further inquiries can be directed to the corresponding author.